View
10
Download
0
Category
Preview:
Citation preview
California State University, San Bernardino California State University, San Bernardino
CSUSB ScholarWorks CSUSB ScholarWorks
Theses Digitization Project John M. Pfau Library
2003
The Euler Line in non-Euclidean geometry The Euler Line in non-Euclidean geometry
Elena Strzheletska
Follow this and additional works at: https://scholarworks.lib.csusb.edu/etd-project
Part of the Mathematics Commons
Recommended Citation Recommended Citation Strzheletska, Elena, "The Euler Line in non-Euclidean geometry" (2003). Theses Digitization Project. 2443. https://scholarworks.lib.csusb.edu/etd-project/2443
This Thesis is brought to you for free and open access by the John M. Pfau Library at CSUSB ScholarWorks. It has been accepted for inclusion in Theses Digitization Project by an authorized administrator of CSUSB ScholarWorks. For more information, please contact scholarworks@csusb.edu.
THE EULER LINE IN NON-EUCLIDEAN GEOMETRY
A Thesis
Presented to the
Faculty of
California State University,
San Bernardino
In Partial Fulfillment
of the Requirements for the Degree
Master of Arts
in
Mathematics
by
Elena Strzheletska
December 2003
THE EULER LINE IN NON-EUCLIDEAN GEOMETRY
A Thesis
Presented to the
Faculty of
California State University,
San Bernardino
by
Elena Strzheletska
December 2003
Approved by:
Robert Stein, Committee Member
Susan Addington, Committee Member
Peter Williams, Chair Department of Mathematics
Terry Hallett,Graduate Coordinator Department of Mathematics
ABSTRACT
In Euclidean geometry, the circumcenter and the centroid of a nonequilateral
triangle determine a line called the Euler line. The orthocenter of the triangle, the point of
intersection of the altitudes, also belongs to this line. The main purpose of this thesis is to
explore the conditions of the existence and the properties of the Euler line of a triangle in
the hyperbolic plane. As a model of hyperbolic geometry Poincare’s conformal disk model
is going to be used. For the algebraic representation of the geometric objects on the
hyperbolic plane, we chose to use Hermitian matrices. In this paper we will find Hermitian
representations of the three important points of a hyperbolic triangle (the circumcenter,
centroid and orthocenter) as well as the Hermitian matrix that represents the Euler line,
and discuss the conditions of their existence. We will complete our discussion by proving,
that if the circumcenter and orthocenter of the hyperbolic triangle exist, the Euler line goes
through the orthocenter if and only if the triangle is isosceles.
iii
ACKNOWLEDGEMENTS
I would like to express my deepest appreciation to Dr. John Sarli for all of his
time, assistance and input in the writing and completion of my thesis. During my graduate
studies, I have found him to be challenging and encouraging, insightful and patient; his
door has always been open. I would like to thank Dr. Robert Stein and Dr. Susan
Addington for their work as members of my thesis committee. In addition, I would like to
express my gratitude to all of my professors at California State University, San Bernardino
as well as those at M.P. Drahomanov National Teacher’s Training University, who helped
me develop a strong foundation in mathematics. Finally, I would like to thahk my husband
for all of the assistance and support he has provided in the completion of my thesis.
iv
TABLE OF CONTENTS
ABSTRACT................................ iii
ACKNOWLEDGEMENTS.................................................................... iv
LIST OF FIGURES........................................................................................................vii
CHAPTER ONE: HISTORY AND BACKGROUND
1.1 History of Non-Euclidean Geometry.................... 1
1.2 Poincare's Version of Hyperbolic Geometry - Conformal Disk Model . 3
1.3 Hermitian Matrices..................................................................................... 5
1.4 Pencils of Circles........................................................................................ 8
1.5 Notations......................................................................................................8
CHAPTER TWO: A HYPERBOLIC TRIANGLE AND ITS CEVIANS
2.1 A Hyperbolic Triangle ............................................................................10
2.2 Perpendicular Bisectors ............................................................................12
2.3 Medians ....................................................................................................15
2.4 Altitudes ................................................................................................... 21
CHAPTER THREE: HERMITIAN REPRESENTATIONS
3.1 Circumcenter ............................................................................................25
3.2 Centroid ....................................................................................................28
3.3 Orthocenter................................................................................................32
CHAPTER FOUR: THE ANGLES OF PARALLELISM
4.1 Howto Define the Kind of Pencil ........................................................... 36
4.2 Circumparallelism ............................................................ 40
v
4.3 Orthoparallelism....................................................................................... 42
CHAPTER FIVE: EULER LINE AND ITS PROPERTIES
5.1 Euler Line ............................................................................ 47
5.2 When does the Orthocenter Belong to the Euler Line ........................... 53
CHAPTER SIX: SUMMARY ....................................................................................59
REFERENCES ............................................................................................................63
vi
LIST OF FIGURES
Figure 2.1........................................................ 14
Figure 2.2............................................................................................ 19
Figure 2.3........................................................................................................................24
Figure 4.1........................................................................................................................36
Figure 4.2......................................................................................................................40
Figure 4.3........................................................................................................................43
Figure 4.4........................................................................................................................46
vii
CHAPTER ONE
HISTORY AND BACKGROUND
1.1 History of Non-Euclidean Geometry
The notion that there exists a geometry different from Euclidean geometry is
relatively new for mathematics. It was not until the 19th century that "a new geometry
contrary to Euclid’s" was found to be "logically possible." [12] Consider the Euclidean
Parallel Postulate:
“If a straight line falling on two straight lines makes the two interior
angles on the same side of it taken together less than two right angles, the two
straight lines, if produced indefinitely, meet on that side on which are the angles
together less than two right angles. ” [9]
This proposition was stated in Euclid’s Elements as the fifth (parallel) postulate.
Modern books on geometry more often use the equivalent of the postulate: Through a
point not on a given line there is a unique straight line parallel to the given line. The
discussion about the parallel postulate lasted over 2000 years. Thinking that the fifth
postulate depended on the other four, scientists continually tried to prove it.
Non-Euclidean geometry was born as soon as the question about the independence of
Euclid’s parallel postulate was resolved.
One of the world’s greatest mathematicians, Karl Friedrich Gauss (1777-1855),
was probably the first (around 1813) to discover the existence of consistent geometry
where Euclid’s postulate was replaced by the contrary statement. To such a geometry,
Gauss gave the name "non-Euelidean". Gauss would eventually be considered the founder
1
of non-Euclidean geometry, even though the German mathematician, physicist and
astronomer, with international reputation, was fearful of being misunderstood, and never
published his discovery. Later on, Ferdinand Schweikart (1780-1859), professor of
jurisprudence, basically came to the same conclusion as Gauss, but also failed to publish
his views. . ? ,
The great Russian mathematician Nicolay Lobachevskiy (1792-1896) approached
the problem in a different way. To show the dependence of the parallel postulate on the
other four axioms, he chose to use the proof by contradiction. His thought was the
assumption that through a given pointnot on a given line could go at least two straight
lines parallel to the given line will lead him to a contradiction. The longer he worked, the
more he believed that there was no contradiction in the new, geometry he discovered.
Having realized that geometry built on the assumption that Euclid’s parallel axiom was not
true does not contain a contradiction, Lobachevsky began to develop a new non-Euclidean
geometry, and made a great impact on its foundation. In 1830 Lobachevsky was the first to
publish a work paper on non-Euclidean geometry, On the Principles of Geoimetiy . Even
though the Hungarian mathematician Janos Bolyai (1802-1860) had actually made the
same discovery of the new geometry before Lobachevsky, his was published later, in 1832,
and only as.an appendix to his father, Farcas Bolyai’s, mathematical book.
That is why, during a 20-year period, non-Euclidean geometry was independently
discovered four times . Thus, history gives credit for the discovery to all four
mathematicians, and this geometry is sometimes called "Gauss-Schweikart-Bolyai-
Lobachevskian," but is most often called "Bolyai -Lobachevskian," giving special
recognition to those scientists who were willing to publish their findings. As with
2
everything new, non-Euclidean geometry was not understood and accepted for some time,
even by advanced mathematicians, and Lobachevsky’s work was even called "a satire" and
"a caricature" of geometry.
The next phase in the evolution of non-Euclidean geometry began with developing
its models. During the second half of the 19th century a number of models of hyperbolic
geometry were introduced by Eugenio Beltrami, Felix Klein and Henri Poincare. (In 1871,
German mathematician Felix Klein proposed to call Euclidean, Bolyai-Lobachevsky and
Riemann geometry " parabolic", "hyperbolic" and "elliptic" respectively). For this paper,
we shall focus on one of the two developed by French mathematician and physicist Henri
Poincare (1854-1912) - his conformal disk model.
1.2 Poincare’s Version of Hyperbolic Geometry - Conformal Disk Model
To obtain this model, we will use a stereographic projection. Let us consider a unit
hyperboloid of two sheets T72, and a unit.circle C that belongs to the generalized
complex plane C . Let also S be the South Pole of H2. Under the stereographic projection
of H2, from, the point S to the generalized complex plane C , the points of the upper and
lower planes of H2 become the points of the interior and exterior of the unit circle
respectively. .The pairs of antipodal points of the upper and lower sheets of H2 become the
pairs of points on C that map to each other under the inversion in £. A section of a plane
that goes through S and has common points with thehyperboloid is a hyperbola. Under
the projection the image of the hyperbola is either an arc of a circle, orthogonal to the unit
circle (if the plane does not contain z- axis), or a straight line that goes through the origini ...
(otherwise). The last point we should make is that the stereographic projection preserves
3
the angles (it is conformal), but changes the distances. Now that we have described how to
obtain Poincare’s conformal disk model, let us look at the model itself.
The points or d -points in Poincare’s model of hyperbolic geometry consist of
those points in the unit disk 2) that do not belong to its boundary. “A aMine is that part of
a (Euclidean) generalized circle, which meets £ at right angles and which lies in 2)”. Two
d-lines are called parallel if the generalized circles that afford them have a common point
on the boundary of D, and ultra-parallel if the generalized circles that afford them do not
meet either in 2) or on £. “A d -triangle consists of three points in the unit disc 2) that do
not lie on a single J-line, together with the segments of the three (Alines joining them”. [1]
We also want to state a few basic definitions, theorems and formulas. To find the
non-Euclidean distance d(z\,Z2) between the points z\ and 22 in the unit disk 2), we use
the formula:
(M) rf(z,,z2) = rt-'(-a^-).
(For ease of the notations, shx, chx, thx, cthx, schx will stand for hyperbolic sine,
hyperbolic cosine, hyperbolic tangent, hyperbolic cotangent and hyperbolic secant
respectively. It follows from (1.1) that the non-Euclidean distance af(0,z) between the
points 0 and z in the unit disk 2) is
(1.2) al(0,z) = th~\\z\y.
Let z\ and zi be two points in the unit disk. There exists a unique J-line through z\
and zi, and the center t] of the circle affording this al-line could be found as follows:
fl 3) n = (Zl -z2)_~zi^2(zi -Z2)t • J */ Z\Z\-ZiZi
4
Let p and q be two distinct points in the unit disk. Let £ be the center of the circle
that affords the perpendicular bisector of the ri-line through p and q. Then
r = (p - q) + pq(p - q)' ■ ' pp ~ qq ’
provided that |p| #= \q|. In the case when [p| - |</| = 0, the perpendicular bisector to the
J-line through p and q is a straight line - the diameter of the unit disk, that lies on the
bisector of the angle pOq.
1.3 Hermitian Matrices
A complex 2x2 matrix H is Hermitian, provided that H = H*, where H* is the
conjugate transpose of H. It follows that
(1-5)A B B D
H =
where A,D e R, and B e C.
Proposition.
The following form allows us to represent a circle in C as the locus of all z e C
such that
(1-6) = ( z = 0,J\ B D 1
from which it follows that the locus is a circle in C provided that detLt < 0.
From the above, we can make the following observations that could be easily
verified:
5
1) Hermitian matrix
(1-7) Hu =1 0 0 -1
represents the unit circle.
2) If A = 0, (1.5) represents a straight line.
3) Let ffl and k < co be respectively the center and the radius of some circle in C.
Let also
A B B D
H =
be a Hermitian matrix that represents this circle. Then
A B B D
d-8) = A1 —Q)
— co com - k2
4) According to the Proposition stated above, two Hermitian matrices H\ and H2
represent the same circle if and only if I~L = kH2 , were A is some nonzero real number. In
our computations we will be interested in Hermitian matrices as representatives of circles,
thus, though strictly speaking, two matrices Hx and such that Hx = ZH2 are not equal,
we won’t make a distinction between them. To stress this real - projective equivalence of
Hermitian matrices, instead of usual parentheses
\BD
we will use square brackets:
6
A B B D
Let H2 denote the Hermitian matrix adjoint to H. The inner product of two
Hermitian matrices and H2 is defined by
(1-9)
where
(MO)
= Tr(H,H2),
Tr(H\H2) = A}D2 +A2D1 — B}B2 — B2B1.
It follows that if Hermitian matrices and H2 represent circles orthogonal to the unit
circle:
\){HX,H2} = 2AXA2-B,B2-B2B,
2) <tfi,tfi) = 2(Aj -Bj) = 2dettfj
Since any circle that affords a d-line is orthogonal to the unit circle, it is
represented by Hermitian matrix
(Ml)A B B A
H =
Let Ci and C2 be two generalized circles, that are represented by Hermitian
matrices Hi and H2, respectively. The angle <p between C\ and C2 we define as follows:
(1-12) cos<p = (Hi,H2)
J(Hi,Hi)-(H2,H2)
We can conjecture from (1.12) that:
7
1) Ci and C2 intersect if and only if |cos <p | < 1,
in particular Ci and C2 are orthogonal if and only if cos(p - 0,
2) Ci and C2 are tangent to each other if and only if |cos (p | = 1,
3) Ci and C2 do not have common points if and only if |cos q> | >1.
1.4 Pencils of Circles
Euclidean geometry considers two kinds of pencils - concurrent and parallel. In
hyperbolic geometry there are three kinds of pencils. In addition to elliptic and parabolic
pencils, which are analogous to the concurrent and parallel Euclidean pencils, respectively,
there exist hyperbolic pencils of circles. Any two distinct circles on the hyperbolic plane
generate the pencil of circles. If Cj and C2 are two circles in C, then any circle C that
belongs to the pencil defined by Cj and C2 can be expressed as a linear combination of the
two generators of the pencil. In other words,
(1.13) C = 2Ci+juC2,
where 2,/x e R.
1.5 Notations
1) We will use the small letterers: h, k, etc. to denote Euclidean measures and
capital letterers: O, H, K to state the vertices or d -points and non-Euclidean measures -
the context will determined which usage is intended.
2) Sometimes, when it is understood from the context, we will call the circles by
the names of J-lines they afford.
8
3) We will use
(1-14)
for the inversion of p in the unit circle.
9
CHAPTER TWO
A HYPERBOLIC TRIANGLE AND ITS CEVIANS
2.1 A Hyperbolic Triangle
Let HOKbe a hyperbolic triangle in the unit disk. Without loss of generality, we
can assume that one of the vertices of the triangle, say O coincides with the origin, and the
side HO is on the real axis. Let also the sides HO and KO have the Euclidean lengths h
and k respectively (h > k), and the angle KOH equal 6. Then the vertices of the triangle
have coordinates: O = 0, H = h, K = keid. The two sides of the triangle KO and HO are
straight lines. They are represented by the Hermitian matrices
Hko
Hho0 i -i 0
0 sin#+ z cos#sin#-zcos# 0
A d-line HK is an arc of a (Euclidean) circle. To find the center tjhk of this circle, we use
(1.3), wherez\ = /zandz2 = keie:
(h - ke'8) - hkei8(h - ke i8) hk(e~'8 - e,e)
= /z(l + F) - fe;9(l + /z2)hk(e~'8 - e’8)
Given that ei8 = cos#+ z sin# and e~'8 = cos# - z'sin#,
_ /z(l + fc2) - fc(cos# + zsin#)( 1 + /z2) ^HK hk(cos Q - z sin 0 - cos 0 - z sin #)
10
fc(l + A2)sin0 + i(h(l + k2) -k(l + h2) cos O') hk(cos6 - isinO - cos 6 - zsin0)
Let
(2.1) k(l + h2) = k+ and h(i + k2) = h+.
Then from the last expression
(2.2) TjHK = k+ sinfl + i(h+ - k+ cos O')
2hksin0
According to (1.8) and (2.2), the expression for the Hermitian matrix of a d -line
that is obtained from the point t]hk can be written in a form:
Hhk =1 -Vvk
-tjhk 1
(2.3) Hhk =1
-&+sin0 - i(h+ — k+cosO) 2hksinQ
-k+ sinfl + z'(/z+ - k+ cost?) 2/z&sin0
1
Multiplying the last matrix by 2hks'm6 e R, we obtain another matrix that by (1.6) also
represents Hhk- And so:
(2.4) Hhk =2hksin8 -k+sin 8 + i(h+- k+cos 0)
-k+ sin8- i(Jj - k+ cos0) 2/z£sin0
In our further calculations we may use Hermitian representations of Hhk of either form
(2.3) or (2.4), depending on which is more convenient.
Now that we have found the Hermitian representations for the sides of the triangle
HKO, we can calculate its angles H and K, but first we need to compute the inner
11
products (Hi,Hi), (H2,H2), and (Hi,H2). By (1.9) and (1.10)
(Hi,Hi) = -2,
(Hz,Hz) = -2((h+)2 + (F)2 -2A+Fcos0-4/22Fsin20,
(Hi,Hz) = -2(/z+-Pcos0).
Hence by (1.12)
(2-5) cos H = -________ (h+ - k+cos8)___________J(h+)2 + (k+)2 - 2h+k+ cosO - 4h2k2 sm29
and
(2-6) cosK = (k+ - h+ cos 0)7(/z+)2 + (k+)2~ 2h+k+ cos 8 - 4h2k2 sin20 ’
2.2 Perpendicular Bisectors
Let Phk, Pho and Pko be perpendicular bisectors to the sides HK,H0 and KO,
respectively. Let also the centers of the circles that afford the perpendicular bisectors be
^hk-^ho^ko, which we can calculate using (1.4):
(fe,e - h) + khel9(ke~'e - h)QhK~ kW-h2
_ ke'd -h + k2h - kh2eld k2-h2
keie([ -h2)-h(\ - k2)k2 - h2
and finally
12I
^HK k2-h2 '
(ho - I
1 o^K0 = IF* = ~T'
In the formula for Qhk we used substitutions:
(2.7) A(1 -£2) = h~ and &(1 - h2) = k~.
Using (1.11) and the properties of Hermitian matrices, we obtain the expressions for Phk,
Pho and Pko-
Hphk =k2 - h2 h - k e *
h - k e‘e k2 - h2
Hpho
Hpno
h -1
k -e * -ete k
Now we shall discuss the question of whether all three perpendicular bisectors
belong to the same pencil of circles. According to (1.13), in order to prove that HpH0,
HpK0 and HpHK are the members of the same pencil we need to find A and // e R such that
(2.8) hHpH0 + p.HpKO = Hphk.
The left hand side of (2.8) :
13
kHpH0 + pHpK0 Zh -1
-1 hk -e ,e
-e‘e k
kh + pk -Z - pe i9 -Z - pe'9 hh + pk
Comparing the last matrix with the matrix for Phk,vq can easily find the values for Z and
p such that (2.10) holds. When Z = -h~, and p = k~, HpHK is a linear combination of
Hpm and HpK0. Hence we have proved the following theorem.
Theorem 2.1.
The perpendicular bisectors of the sides of a hyperbolic triangle belong to the
same pencil of circles.
Corollary
The circumcenter of a hyperbolic triangle exists if and only if the pencil of the
perpendicular bisectors is elliptic. Figure (2.1) shows LHKO, whose perpendicular
bisectors are concurrent in the circumcenter of the triangle.
Figure 2.1
14
2.3 Medians
Let Mh, Mk and Mo be the non-Euclidean midpoints of KO, HO and HK,
respectively. Since the Euclidean distance between the points Hand O is h and the
non-Euclidean distance is Hit follows from (1.2), that
(2.9a) h = thH,
and similarly
(2.96) k=thK,
We need to express the distance m? between the origin and the non-Euclidean midpoint
Mh in Euclidean terms. The non-Euclidean distance between the O and Mh is ±-H,
hence
1 - Jl-th1 2H = ThH
1 - 7l -62 h
Thus
mHO = -^-(1 - 7l -h2).
Using similar arguments and considering the fact that the angle between KO and
the real axis is 6, we find
mKo = j(l - Jl-k2 )e'e.
15
It would be much more complicated to calculate the third midpoint mmc, but
fortunately there is no need to do it, and the reason for it we will explain later in this
chapter . Knowing the expressions for mno and mKo, we can write the equations of two
medians HMh and KMk, and then find OMo as a linear combination of the first two.
We will start from finding t]kmk, the center of the circle that affords J-line KMk-
To do that, we use the formula (1.3), where z\ = K = ke‘e, and
Z2 = Mk = 4-(l - -h2). Thenn
0_ 1-TT^F _ _ i-/r^„KM =-________ h______________ h___________________h _L
k(e'6 - e~i0) (1 - A~h2 )h '
h2ke‘e - - V1 - ) - 0 - 71 -tifrke'0(hke~iQ - (l - Vl - /z2 ))
h(l - Jl-h2 )k(e,e - e~10)
kei0(h2 + (l - V1 - ft2 ) ) - h(l - Vl - h2 )(1 + k2)
h(l - ~h2 )k(e'e - e~10)
2kei0(l -y/l-h2)- A+(l - Vl - A2) h(l — - h2 )k(e10 - e-'0)
2ke'9 - h+hk(e,e - e~10)
Multiplying by i the numerator and the denominator of the last expression, and recalling
that e‘e = cos0 + /sin0, we obtain:
IjKMg =2£sinfl + i(2kcos9- h+)
2hksmQ
16
Then
HkMk =72Msin0 ,-2£sin0 + i(JP - 2£cos0)
-2£sin0 - i(h+ - 2kcosd) 2hksind
Performing similar calculations, we find
£+ sin# + i(2h - k+ cosd)T]HMh = 2hksirid
which leads to
Hhmh —2hksmd -k+sind + i(2h - k+cosd)
-k+ sin d - i(2h - k+ cos d) 2hk sin d
Now we will return to the question of the third median. Let us first state a theorem.
• Theorem 2.2
The medians Of a hyperbolic triangle belong to the same pencil of circles. The
pencil is elliptic.
Before we start proving this theorem, some preliminary work should be done.
Definition 2.1
If A, B and X are distinct points on a d-line, then their hyperbolic ratio is
h(A,X,B) = » ifAfis between^ and B, andsn\a(X,B))
otherwise [13].
Lemma 2.1 (Converse of Ceva ’ s Theorem)
Let ABC be a hyperbolic triangle, and P, Q, R three points in the unit disk. If P lies
on the tf-line AB, Q on BC and R on CA in such a way that
17
h(A,P,B) • h(B,Q,C) • h(C,R,A) = 1,
and two of the d -lines CP, BR and AQ meet, then all three are concurrent [13].
Lemma 2.2
Let l\ and Z2 be two circles, orthogonal to the unit circle. Then any generalized
circle of the pencil generated by Zi and Z2 is also orthogonal to the unit circle.
Proof. Let Hi, Hi be the two Hermitian matrices that represent l\ and Z2. Since l\
and 12 are orthogonal to C, Hi and Hi have the following forms:
Ai Bi A2 B2and H2 =
_ Bi Ai B2 A2
Any circle from the pencil that is generated by Hi, Hi is of the form
kHi + pH2,
where A, p e R.
kHi + /z//2 — AAi Bi A2 B2+ pBi Ai B2 A2
kA 1 + pA 2 kB 1 + pB2 kBi + pB2 kA 1 + pA 2
and the Lemma is proved since the last matrix represents the circle orthogonal to the unit
circle. Now we can prove Theorem 2.2.
Proof. Let us use triangle HKO as an example of a hyperbolic triangle. Consider
the pencil of circles generated by the two circles affording HMh and KMk ■ HMh and
KMk have a common point (say point M) inside of the unit disk, which leads us to two
18
conclusions. First, by the Lemma 2.1, all three medians are concurrent. Second, the pencil
is elliptic, and the only straight line of the pencil goes through M. By Lemma 2.2, the
straight line is a d-line, and hence goes through the origin. That, means that OMo, the third
median of the triangle, is coincide with OM, the straight line of the pencil, and thus
belongs to it, which proves the Theorem.
Corollary
The.three medians of a hyperbolic triangle are concurrent. You can see the
illustration to the Corollary in Figure (2.2).
It follows from Theorem 2.2 that LHkmk + pHhmh = Hqm0 for some A and p.
Since OMo is a straight line that goes through the origin, it is represented by the matrix
Homo °f the form:
„ r" b ■ .""'I B 0 ■
Now, from the equation
19
2hksinQ -2ksin 0 + i(h+- 2kcos 6)-2ksin0 - i(h+ — 2kcos0) 2hksind
+
+ /z2/z£sin#
-k+ sin# - i(2h - k+ cos O')■k+ sin 6 + i(2h - k+ cos #)
2/z^sin#
0 B B 0
the following conclusions can be made:
1) (A + fi)2hksin6 = 0 and
2) Z(-2A:sin# - i(h+ - 2kcos6)) + n(-k+ sin# + i(2h - k+cosd)~) = B.
The first equation immediately yields fi = -k, and substituting this into the second we
obtain
Z(-2&sin# - i(h+— 2£cos#)) - Z(-£+sin# + i(2h - k+cosO)) = B.
Then
B = A(-2A:sin# + i(h+ - 2kcosQ} + Usin# - i(2h - &+cos0))
= -2A:sin# + A:+sin# + i(h+ - 2kcos# - 2h + k+ cos6)
= -£sin#(l - h2) + z(-£cos#(l - h2) - h(\ - k2)) — -k~ sin# + i(k~ cos# + h~)
Therefore the third median OMo is represented by the matrix
0 -k~ sin # + i(h~ + k~ cos #)Homo =
-k sin#-z'(/z + k cos#) 0
20
2.4 Altitudes
Let Ah, Ak and Ao be the feet of the altitudes of triangle HKO from the vertices H,
K and O, respectively. Our goal is to find the Hermitian matrices that represent the
altitudes. We are going to start from OAo, which is a,straight line and hence its matrix
Hoao is of the form
HoAq =0 B B 0
where B e C. Let B = bi + ibi = ReB + ImB.
Since OAo is an altitude oiHK, OAo goes through the center tjhk of the circle
HK, and thus B has to satisfy the following equation:
Then multiplying through the matrices of the last expression and substituting B
for hi + ibz, we obtain the equivalent equation
(hi-z'h2)'zr//K+(hi+/h2)w =
which simplifies to
hi Re(z/w) - h2 Im(z7tfjO = 0,
If we now recall that
Re('"";) = 737 and h+ - fc+cosfl2/z£sin0 ’
then we receive the equation
21
l fc+sin0 _ r Zz+ - k+ cos 6 = n 1 2A&sin0 2 2/z£sin0
It follows then that
bx h+ — k+cosd j. AT sin 0 °2
We shall use the fact that OA0 ± HK one more time. As it follows from (1.9), the
inner product of two orthogonal circles is zero, so
(Hhk,Hoao) = 0-
The last condition is equivalent to the following equation:
(Usin0 - i(h+ - fc+cos0)) (1 + b2) 2h + ibz
(1 +h2) 2h - ibz )(.k+sind + i(h+- k+cosd')') = 0,
which gives us the solution for Z)2:
b2 =-(1 + A2)A:+sin0 2h(h+ - k+ cos 0)
Then
bi = -(1 +h2) 2h
and finally,
B-(1 +-h2)
2h(h+ - k+ cos 0) ((h+ - k+cosQ) + ik+ sin0).
Using the properties of Hermitian matrices, we write Hqa0 as follows:
22
HqAo
-26cos0((62 + 1)cos0 + z(62 + 1) sin0
HhAh. *
and
-2£cos0 k2+ 1
Hkak =
0 (h+ — k+cosO) + ik+sinQ(h+ - k+ cos 6) - ik+ sin 0 0
Now, applying the condition for HAh and KO to be orthogonal, and for KAk and
HO to be orthogonal, and performing the calculations very similar to the case with
OAo, we obtain:
((62 + 1)cos0 - i(h2 + 1) sin0 -26cos0
k2 +1 -2kcos0
As in the case with the perpendicular bisectors, we want to know if all the three
altitudes belong to the same pencil or in other words is there such A and p e R, that
(2.10) IHhah f pHkak = Hqa02
Comparing the three expressions for the matrices of the altitudes, we observe that
for (2.10) to be true, it is necessary that
1) kh + pk = 0
and
2) k((h2 + l)cos0-z(62 + l)sin0) + p(k2 + 1)
= (hi - k+cos6) + ik+sind. .
Solving the two equations simultaneously, we Obtain the answer: if p = h and T = -k,
then (2.10) holds. Thus we just proved another theorem.
23
Theorem 2.3. ■I
The altitudes of a hyperbolic triangle belong to the same pencil of circles.
Corollary
The orthocenter of a hyperbolic triangle exists if and only if the pencil of the
altitudes is elliptic. In figure (2.3) we have drawn £\HKO, whose altitudes are concurrent
in the orthoeenter of the triangle.
24
— (ai cos# + a2 sin#) = 0.
Now we express a2 through ai to get:
a\(k-hcosd) a2 = —hM----- ’
and substitute this into (3.3):
(“U«5(^A^)2 + i>-2«,=o,
(a2(/z2sin2# + k2 + /z2cos20 - 2hkcos6') + /z2sin2#)/z - 2ai/z2sin2# h2 sin2#
We notice that h2 sin2# =# 0 and can replace the previous equation by its equivalent:
(3.6) a2(h2 + k2 - IhkcosQ) - 2aihsm26 + A2sin2# = 0.
Let us consider only the triangles for which h2 + k2 - 2hkcos0 + 0 (the case
h2 + k2 - 2Mcos# = 0 corresponds to the isosceles right triangle), and write (3.6) in
non-EucIidean terms:
a2(th2H+ th2K- 2thHthKcos#) - 2a\thHsin2d + th2H sin2# = 0.
To simplify the notation, let
(3.7) th2H + th2K-2thHthKco$Q =
and then we can rewrite the last equation as follows:
a]fa(H,K) - 2a\thHsvn2Q + th2H sin2# = 0.
The last is a quadratic equation that we solve using quadratic formula:
thHsin2Q ± thH|sin#| ^sin2# -fa(H,K)01 = f«(fW ’
26
ai =Z/z7/sin0[sin0 ± ^sin20 - fa(H,K) J
fa(H,K) ’
since 0 < d < n. Considering that «i should belong to the unit disk (| «i | < 1) , we have
the only one solution for ai :
thHsin d £ sin d - ^sin20 - fa(H,K) j ai = f^K) ’
Then
(thK - thH cos 0) £ sin d - .Jsin20 -fa(H,K) J ai= MHK)
and therefore
(3.8) -a£sin 0-7sin20-/„(2T;K)]
fa(H,K) (thH sind + i(thK- thH cosd)).
To simplify the notation of (3.8), let
«o = thHsmd + i(thK-thH cosd).
Then
|«0]2 = th2H + th2K-2thHthKcosd = fa(H,K)
and
I”sin0 - ^sin2^ - Joco |2 "I a = —--------------- ------------ -
l«o|2
Finally, applying (1.14), we obtain:
(3.9) a = j^sinf? - .Jsin20 - ]«o|2
27
From the calculations above we excluded the case when fa(H,K) = 0, in which
a = -^(l +/).
Now that we know how to calculate the circumcenter of a triangle, let us state the
following theorem.
Theorem 3.1.
If two perpendicular bisectors of the sides of a hyperbolic triangle are concurrent,
then the point of their concurrence - the circumcenter of the triangle - is the center of the
circumscribed circle.
The analytic proof of the theorem can be found in [13]. We could give an
algebraic proof, which would consists of computing the distances between the
circumcenter and the vertices of the triangle. After performing straightforward
calculations, one can verify, that d(a, O) = d(a, H) = d(a, K) = R, where R is the
non-Euclidean circumradius. We will limit ourselves to finding the non-Euclidean distance
d(a,O') between the points a and O. From (1.2) and (3.9)
d(a,O) — |a| = th~[ ^sin0 - ^sin20 - |«o|2 j|«o l) •
Therefore, the radius R of the circle circumscribed about a hyperbolic triangle is
R = th~x (£sin0 - ^/sin2# - |«o|2 |«oQ•
3.2 Centroid
Let [) = pi + ip2 denote the centroid of a hyperbolic triangle. To find the centroid,
we will use similar arguments that we used to obtain the formula for the circumcenter, that
28
is, find the centroid as a point of intersection of two medians HMij and OMo- Any point of
the extended complex plane that belongs to the circle HMh, has to satisfy the equation:
(/<>) 2Msin0 -&+sin0 + i(2h - h+cos0)-k+sinQ -i(2h - k+oosQ) 2hksind
P = 0.
On the other hand, any point of C that is on OMo satisfies the equation:
{/>>) 0 -k svnQ + i(h +k cos0)-k~ sin 0 - i(h~ + k~ cos 0) 0
P1
= 0
By simplifying these, two equations we get:
(3.10) hksinQ(fi2 + P2 + 1) - (&+sin0ai + (2h - k+ cosQ/Pz) - 0
and
k P\ sin0- (h + k cos6)P2 ’ 0.
From the last equation;
a __ k sin0 a h-i k-cose11'
After we simplify the equation (3.10) and substitute the expression for fh into it, we
receive:
<3I1> (wCT)2«r)2 + (iT+2W-C0S(,)-
By employing the substitution .
y h +k cos0 ’
29
we simplify the equation (3.11):
y2 ((/z )2 + (k )2 + 2k h cos#) -y(( 1 + Zz2)(l - F) + 2(1 - h2)) + 1=0.
After this point it is going to be more convenient to use non-Euclidean distances
instead of Euclidean. So, with the notation (2.9), we rewrite the last equation as:
th2 H th2 Kch4K ch4H
thH thK ch2K ch2H
cos# -y 1 + f/z2 # ch2K ch2H J
) + l=0,+ + 2 +
sh2(2H) + sh2(2K) + 2 sh(2H)sh(2K) cos# * V4ch4H ch4K
&
-ych(2H) + ch(2K) + 1
ch2H ch2K+ 1=0.
Let us introduce the substitutions that will be used throughout the paper:
(3.12) ch(2H) + ch(2K) + 1 =m
(3.13) sh2(2H) + sh2(2K) + 2sh(2H)sh(2K)cosd = fB(2H,2K)
With (3.12) and (3.13) we can rewrite our equation as follows:
r M2H.2K) \ a—vV Ich*H ch*K J V ch'H ch2K J
which (since chH • chK * 0) yields the quadratic equation
y2fB(2H,2K')-y(fm ch2Hch2K) + 4ch4Hch4K = 0.
This equation has two solutions:
--------M1H.2K)-----------2ch1Hcli1K.
30
Let us now express (i in non-Euclidean terms:
a (f thH thK Q. . thK .P = y ((-— „ + —7777- cos# ) +1 sin#),ch2K ch2H ch2H
y(sh(2H) + sh(2K) cos#) + z sh(2K)sin#
2ch2H ch2K
Therefore, the two solutions of (3.11) are:
m - Jm2 -fB(2H,2K) r, x(3-14) jS(i) =------- fB(2H,2K)------------ [Qh(2H)+sh(2K) cos#) + z^(27C)]
and
m +Jm2 -fy(2H,2K) r . xZ><2> = ------- M2H.2K)--------- " ■[(MIW+MVC) cos9) + ish(2K)].
Here we need to discuss whether both P^ and fig) are valid solutions. Using the
properties of hyperbolic functions we show that
fB(2H,2K) = sh2(2H) + sh2(2K) + 2sh(2H)sh(2K)cos0
< ch2(2H) + ch2(2K) + 2ch(2H)ch(2K)
= (ch(2H) + ch(2K)y,
and hence /b(2//,2A?) < m2.
Then
2 (m2 + mjm2 -fB(2H,2K) ) -fB(2H, 2K)
/b(2H,2K)
31
2^m2 +m jm2 -fB(2H,2K) )
fB(2H,2K) -1
2m2fB(2H,2K)
> - 1
= I, '
which shows that the modulus of /?p) is greater than 1, and therefore j3(2) does not belong
to the unit disk. So, ) is the only valid expression for the centroid.
Let
j8o = (sh(2H) + sh(2K) cos 0) + i sh(2K) sin 0.
Then
|/?o I2 = sh2(2H) + sh2(2K) + 2sh(2H)sh(2K) = fB(2H, 2K),
and now we can rewrite (3.14) as:
o m - jm2 - \pQ\2 o P — ------------------- 3------------- po-1/8.012
Using (1.14), we can write the the expression for ft as follows:
(3-15) p=[m-Jm2-\p0\2
3.3 Orthocenter
We start from the observing that if HKO is a right triangle, its orthocenter
coincides with the origin, and hence we can exclude the right triangles from our further
32
discussion. Let y = y 1 + 72 be the orthocenter of a hyperbolic triangle. Using the same
reasoning as in the cases with the circumcenter and the centroid, we find y as a common
point of the two altitudes KAk and OAo. So we need to solve simultaneously the two
following equations:
(r.) -26cos0 £2 + 1
k2 + 1 -2k cos0
71
= 0
and
(-’) 0 (6+ - k+ cos 0) + ik+ sin 0(6+ - 6+cos0) - z'6+sin0 0
These two equations are equivalent to:
(3.16) - 26cos0(/2 + y2 + 1) + 2(k2 + l)y 1 = 0
and
(6+ - k+ cos 0)y 1 - k+ sin 0/2 = 0.
From the last equation
72(6+ - 6+cos0)
U~sin0
After the substitution this expression for 72 into (3.16), we obtain:
,2 , ( h+ - k+ cos 07i +7i( — ^siny") + 1VCOS0“^2 +^7i = 0(7? + 7?(-
71 y Vsin0 J ((F sin0)2 + (6+)2 + (6+cos0)2 26+6+cos0)6cos0 -
- n(F + l)(A+sin0) + kcosd = 0. k+ sm 0
33
With the help of another substitution
nx - k+ sin#
and some simplification we will receive the quadratic equation:
(3.17) x2((F)2 + (/z+)2 -2/z+Fcos#) -x(A:2 + l)(/z2 + l)tan# +l =0.
Let us now introduce two new variables p and q such that
P~ 1+h2 «- 1+t2-
Hence
£2 + l = /z2+l = ^-; k+ = 2A-h+=2hkv r r 7
Then we can rewrite (3.17) as follows:
-2((T)2 + (T)2 - ^eos«) -x(f f tane) + , . 0,
(^2+7?2-2/?^cos#)-2^x-^^-^tan#+1 =0.
The solutions of this equation are:
tan# ± ^tan2# - (q2 + p2 - 2pqcos0) pq q2 + p2 - 2pq cos 0 kkh ’
and thus the solutions for 71:
tan# ± Jtan2# - (q2 +p2 - 2pqcosd)Y1 =------------ 5----- 5—-------- 7-------------q sin#.q2 +p2 - 2pqcos9
We want to express 71 in non-Euclidean terms. But before we do that let us recall
that
34
(3.18) th(2H) = 2thH = 2h 1 + th2H 1 + h2 = p and th(2K) = 2thK = 2k
1 + th2K 1 + k2
and make one more substitution :
th\2H) + th2(2K) - 2th(2H)th(2K) cosd = fr(2H,2K).
Now
7itan0 ± ^tan20 -fr(2H, 2K)
fr(2H,2K) th(2K)sin d,
72tan0 ± ^tan20 -f7(2H,2K)
fr(2H,2K) (th(2H) - th(2K) cosd)
tan0 ± ^tan20 -fr(2H,2K)fr(2H,2K) (th(2K)sin0 + i(th(2H) - th(2K)cosd).7
= q
Let
7o = th(2K) sin0 + i(th(2H) - th(2K) cosd)
Then
|Zol2 = fr(2H,2K)
and
(3.19) 7 = £tan0 ± ^tan20 - |7o|2
In order for 7 belong to the unit disk, we should require |7| < 1. If we perform
some algebraic transformations, we find that if 0 (3.19) we must choose the
negative square root, and if 0 e (-y ;jt) - positive.
35
CHAPTER FOUR
THE ANGLES OF PARALLELISM
4.1 Howto Define the Kind of Pencil
Theorem 4.1
Let Vand W be two circles orthogonal to the unit circle. Let also v = vi + iv2
and w = wi + iw2 be the centers of V and W respectively. Then the pencil of circles,
generated by Vand JV is:
elliptic, if (V1W2 - V2W1)2 > (wi - vj )2 + (1+2 - v2)2,
parabolic, if (viw2 - v2mi)2 = (mi - vi)2 + (m2 - V2)2,
hyperbolic, otherwise.
Figure 4.1
36
Proof. A line, say I, that goes through the points v and w, is the only straight line
of the pencil defined by the circles V and W. We can define the kind of the pencil
generated by V and IT by the position of the line I with respect to the unit circle: if I is
exterior to the unit circle, then the pencil is elliptic (as we have drawn on Figure (4.1)); if
I is tangent to the unit circle, the pencil is parabolic; and finally, if I is interior to the unit
circle, the pencil is hyperbolic.
Now, using the conjecture from (1.12), we can restate the statement above as
follows: let (p be an angle between the straight line I and the unit circle. Then the pencil
generated by V and W is
elliptic, if and only if |cos <p| >1,
parabolic, if and only if |cos (p\ = 1,
hyperbolic, if and only if |cos q>\ < 1.
Therefore our nearest goal is to express cos(p through vandw. In the formula (1.9) let
Hi = Hu represent a unit circle, so by (1.7)
and let also H2 = H be a straight line through v and w. Thus H is of the form
0 B B D
H =
where B = bi + ib2 as usual. Applying (1.9) and (1.10) to our case, we get
Tr(Hu,H) = D,
and thus
37
DCOS(p = ,_______ .-2yjaFtf|
Now we need to express the last formula through the coordinates of the centers of
the two given circles. Since our straight line is supposed to go through both v and w, its
matrix has to satisfy the two conditions:
0 B V
B D 1
and
0 B wB D 1
This gives us two equations:
(4.1) 2(Z>i vi - 62V2) + D = 0 and 2(6iWi - &2W2) + D = 0,
which we need to solve simultaneously. The solutions of (4.1) are:
B =, (W2 ~ V2) + (Wl ~ Vl)/
W2-V2
and
D = 2b V2W1 - V1W2W2 ~ V2
Thus Hermitian matrix H; of the straight line I is:
Hlbi
(W2 - V2) + (wi - Vl)i W2 - V2
(w2 - V2) - 0*1 - Vl)/
W2 ~ V2 2b V2H’i - V1W2W2 ~ V2b
0
Using (1.6), we can rewrite Hi as follows:
38
Hi =0 (w2 - V2) + (wi - Vi)z
(W2 - V2) - (wj - Vi )z 2(V2Wl-VlW2)
The determinant of the last matrix is:
detHz = - V2)2 + (wi -n)2).
Finally we find the angle (p between the straight line I and the unit circle as:
(4.2) COS (p = V1W2~V2Hi1
7(^i - vi)2 + (w2 -V2)2
It then follows that the pencil is elliptic, if and only if
V1W2 - V2W1
7(wi - Vl)2 + (w2 - V2)2 ,
parabolic, if and only if
V]W2 - V2W1■ 7(wi - Vl)2 + (w2 -V2)2 .
and hyperbolic, if and only if
' V1W2 —V2W11 X X 0 X s')7(wi - V1)2 + (w2 - V2)2
which proves Theorem 4.1.
39
4.2 Circumparallelism
Definition 4.1
Let us consider a set Phk of hyperbolic triangles HKO with the sides KO and HO
of fixed length H and K respectively, and the angle HOK = Q, where 0 < 6 < n. The
angle 6P correspondent to the a HKO e Phk, whose perpendicular bisectors are mutually
parallelwe will call the angle of circumparallelism. In Figure (4.2) we demonstrated
a HKO whose angle 6 is the angle of circumparallelism.
Figure 4.2
It was proved earlier that the circles that afford the three perpendicular bisectors of
the sides of a triangle belong to the same pencil. The point of their concurrence exists and
belongs to the unit disk only in the case when the pencil is elliptic, thus we can not claim
that the circumcenter of the hyperbolic triangle always exists. To explore the conditions
of its existence, according to Theorem 4.1, we need to find the angle (pp between the only
40
straight line of the pencil and the unit circle. Consider the circles KO and HO to be
generators of the pencil of perpendicular bisectors. Let the centers of KO and HO be v
and w respectively. Thenv = = -^cos# +z-^-sin# and w = Jk Substituting y and
w into (4.2), and performing some algebraic transformations, we obtain:
-sin# -sin# -sin#cos(pp - ■■ ........ '■ ■ — = •---- - = —:------ .Jh2+ k2-2hkcos6 Jth2H + th2K-2thHthKcos9 («o [ -
Hence the perpendicular bisectors are parallel if
sin2# 1th2H + th2K-2thHthKcosO ~ ’
Given that sin2# = 1 - cos2#, the previous equation is equivalent to the quadratic
equation
cos2#- 2thHthKcos9 +(th2H + th2K- l) = 0,
which yields
cos# = thHthK + /(l - th2H)(i - th2K),
(4.3) cos# = thHthK + schHschK.
Let us notice here that since we agreed that H > K,
- 1 < thHthK - schHschK < 0 and 0 < thHthK + schHschK < I ,
and so if the angle # of the triangle HKO is acute, its angle of circumparallelism is>
cos Qp = thHthK - schHschK,
while if the angle # is obtuse, the angle of circumparallelism is
41
cos 6P = thHthK + schHschK.
It follows, that the pencil of the perpendicular bisectors is elliptic if and only if
(4.4) thHthK - schHschK < cos0 < thHthK + schHschK
In the case of an isosceles triangle,
cos 8 = th2H±sch2H,
and since the case cos0 = th2H+ sch2H = 1 is impossible for the triangle, we are left
with only one angle of circumparallelism:
cos8p = th2H - sch2H,
which simplifies to
(4.5) cosdp = 2th2H- 1.
Another way to express (4.5) is:
cos -y- = thH,
In this case the pencil is elliptic if and only if
(4.6) cos8 > 2th2H- 1.
The conditions (4.4) and (4.6) are also the conditions of the existence of the circumcenter
of a hyperbolic triangle.
4.3 Orthoparallelism
Definition 4.2.
Let us consider the set Phk of hyperbolic triangles, that we defined above. The
angle 8a corresponding to the a HKO e Phk, whose altitudes are mutually parallel we
42
will call the angle of orthoparallelism. In Figure ( 4.3) three altitudes of a HKO meet on
the boundary of the unit circle and thus mutually parallel.
Figure 4.3
Let <pa denote the angle between the only straight line of the pencil of the altitudes
and the unit circle. Using arguments similar to those in the case with the perpendicular
bisectors, we calculate cos<pa :
. , A+A+ sin#cos®a = ------------ , —- , -2AAcos9jh+ + k+ - 2hkcosQ
The condition for the altitudes to be parallel is |cos(pa\ = 1 or cos2<pa = 1.
Squaring the last equation and applying the substitutions (3.18) from the Chapter III, we
obtain the following equation:
P+
pqcos#)
COS2#+ 1=0,
which yields the cubic equation
43
(4.7) 2pqcos39 - (p2 + q2 + l)cos20 + 1 = 0.
The last equation has three real roots, but only two of them have modulus less then one:
cos0(i) e (-1,0)
cos0(2) e (0,1),
and thus we have two angles of orthoparallelism - one acute, and one obtuse.
In the case of an isosceles triangle (4.7) becomes
2p2 cos30 - (2p2 + 1) cos20 +1=0,
(cos0 - l)(2/?2cos20 - cos0 - 1) = 0.
Since cos0 * 1 for any 0 e (0,tt), the only possibility we have is
2p2 cos20 - cos0 -1 = 0,
solutions of which are:
COS0 =1 ±7l + 8^2
4p2
Let us evaluate the expression1 + 7l + 8p2
4p2. Since p = th(2Hj < 1,
1 + 7l + 8p2 p + 7p2 + 8p2 1----- J---------- > ----- 1------------ = — > }4p2 4/?2 P
Since we need |cos0| < 1, in the isosceles case there is one angle of orthoparallelism:
1 - 7l + Sp2 COS da = -----------—T---------,
4p2
or after we substitute p = th(2H),
44
(4.8) COS#a1 - 7l + 8//z2(277)
4//z2(2H)
The function cos#a(th(2H)) does not have critical points, and therefore it varies
monotonically with H. It follows that the pencil of the altitudes is elliptic, if 0 < 0 < #a,
or
(4.9)1 - 7l + 8th2(2H)
4th2 (2.H)
The next question we want to discuss - is there any isosceles triangle such that its
angle of circumparallelism is also the angle of orthoparallelism? It follows from (4.5) and
(4.8), that this is true if
1 - 7l + 8t/z2(2H) 4?/z2(277)
= 2th2 H- 1.
The solution of the last equation is
th(2H) =
and thus there is a unique hyperbolic triangle, that has the required property. Its angle 0 of
ortho- circumparallelism is obtuse and cos# = —y, the length of the equal sides is
fs /?H = th 1~5—, and the base HK is of the length HK = th 1 —— ■ In Figure (4.4) we have
drawn a HKO whose angle of circumparallelism is also the angle of orthoparallelism.
45
Figure 4.4
Finally, combining (4.6) and (4.9), we obtain
Qa, if A <cos# > max [#a ,#?] =
dp, ifh >
which states the conditions that the angle # of the isosceles hyperbolic triangle needs to
satisfy in order for both the circumcenter and the orthocenter exist.
46
CHAPTER FIVE
EULER LINE AND ITS PROPERTIES
5.1 Euler Line
Definition 5.1.
A hyperbolic Euler line is a d-line that goes through the circumcenter and the
centroid of a nonequilateral hyperbolic triangle. We should add two remarks to this
definition:
1) The Euler line of a hyperbolic triangle exists if and only if there exist the
circumcenter and the centroid.
2) In an equilateral hyperbolic triangle (as in an equilateral Euclidean triangle) the
circumcenter and the centroid coincide (the proof of this property can be found in a
number of books on non-Euclidean geometry), and define a point-circle instead of a
d -line.
Let us recall one of the expressions of the hyperbolic circumcenter that we
obtained in Chapter III:
a = l«o|2
where ao = thHsinQ + i(thK- thHcos#). To simplify the notations in our further
calculations, let us introduce the following substitutions:
(5.1)
aoi = thHsm.9 and <202 = thK - thH cosO.
47
Then «o = «oi + zao2 and
(5-2) - -r^-(aoi + *<*02) l«o|2 ■
a =
a a 10|ac
(aoi +/ao2)
Now, in the expression of the centroid:
o m -Jm2 -\p0\2 op ■ —Ji—p°’
where po = (sh(2H) + sh(2K) cos 8) + i sh(2K) sin0, let
(5.3) pio = [m- Jm2 - \po\2 J,
Poi = sh(2H) + sh(2Kj cos 8 and po2 = sh(2K)sind.
It then follows that Po = poi + 1P02 and
(5.4) (ft.+ife).I/J0I2
For the algebraic representation of the Euler line, let us use the Hermitian matrix
He of the form:
(5.5)1 B B 1
He =
where B = b\ + ib2. It follows from (5.5) and the definition of the Euler line, that
(5-6)1 B B 1
a(“ 1) -0,
48
1 B B 1
P(5.7) (**) = 0.
Then from the equation (5.6) we will have:
(a2 + «2 + 1) + 2(b\ai -620:2) = 0,
and if we now recall that aj +a2 = |a|, we can rewrite the last equation as follows:
(5.8) |a|2 + 2(6iai - b2a2) + 1 = 0.
According to (5.2), |oc|2 = Tlun anda2ol«o|2
6jai -b2a2 Q10M (/»1 CKoi - />2«02).
Applying this information and observing that |ao| * 0, from the equation (5.8) we obtain:
afy+2aio(aOi6i - 0:0262) + |a0|2 = 0,
which yields the expression for b2 :
(5-9) = aio + la°|2 + 2-awaoibi2o!ioQ!02Now, using arguments similar to the case with the circumcenter, and applying
(5.4) to (5.7), we obtain
Pm + 2.Pm(Po\b\ —Po2b2) + |/?o|2 = 0.
By substituting the value of b2 from (5.9) into this equation we receive:
/?20o: 100:02 + 2y3 iojSqi61 1OC02 - P10P02(o:2o + |«o|2 + 2a io«oi6i) + |/?o |2ce io<*o2 — 0-
Therefore
49
(5.10)fiio/M^io + |«o|2) - aioQQ2(ftiO + l/?o |2)
2(jSioj3oiaio«o2 - aioaoi/?io/?02)
According to (5. l)a20 = 2sin20 - |a012 - 2sin07sin20 ~ |«o|2 ,
«20 + |«o|2 = 2a 10 sin 0.
It follows from (5.3) that
Pv> = 2mpw — |/?0|2.
Pw + l/W = 2mPi0
Now, we can rewrite (5.10) as follows:
_ pQ2 sin0 - gym 1 j3oi«O2 - «oi)5o2 ’
Then from (5.9) and (5.10) we find 02 :
= sin0 + apibi _ Poi sin0 - apizu 2 a°2 P01OC02 - (X01P02
Let
(5.11) A = Poiao2 - aoiPo2-
(It is interesting to notice that /?oi«o2 - ao\Po2 = A = Im(ao/Io).) With the last
substitution made,
B = -^-((/Io2sin0 - amm) + z'(j0Oisin0 - aoim))
= -i- sin0(jSO2 + iPoi) - ^(«02 + /«oi)
(P02 sin0 - ao2w) + i(Po\ sin# ~ aoi«?)A
50
Finally, the Hermitian representation He for the hyperbolic Euler line is found:
He =
(jS02 sin0 - a^m) - z'(/?oi sin0 - aoi«?)
(/?O2sin0 - ao2Z«) + z()Soi sin0 - aoi»z) A1
1
(5.12) He =
A (Pm sin0 - ao2Z«) + z(jSoi sin0 - aoizw)(/?o2sin0 - ao2»z) - i(Poi sin0 - aoizw) A
We could simplify the notation if we notice that
B = -L sin0(/?o2 + ifioi) - zw(a02 + zaOi)
= -=4sin0(j0Oi - z‘/?02) - zw(aOi - ^02))
= -^-(zwao - /?osin0).
The last observation gives us another form of He :
He =A i(mao - /?osin0)
-i(mao - )5osin0) A
where A = Im(ao j0o).
As a verification of our expression, let us find Hermitian representation of the
Euler line in the isosceles triangle. Consider the nonequilateral triangle HKO such that
HO = KO = H. To compute A and B1 = A • B, we use (5.11) and (5.12) :
A - sh(2H)(l + cos0)//zH(l - cos0) - shQH)thHsm2Q = 0.
51
B' = ()3o2sin# - ao2w)+z(j3oi sin# - aoizw)-
Re(R') = sh(2H)sin29 - thH(1 - cos#)(2c/z(2tf) + 1)
= ZA//(2 cA2/7sin20 - (1 - cos#)(2sA2//+ 2 cA2H+1))
= thH(2 ch2H(sin29 - (1 - cos#)) - (1 - cos9)(2sh2H+ 1)
= thH (2 ch2Hcos9 - 2sh2H- l)(l - cos#)
= \thH(2ch2Hcos9 - 2sh2H- l)]2sin2-^-.
Im(R') = sh(2H)(\ + cos 9) sin 9 - thH sin 9 (2ch(2H) + 1)
= ZA7/sin#(2cA27/(l + cos#) - (2sh2H+ 2ch2H+ 1)
= thH(2ch2Hcos9 - 2sh2H- 1) sin#
= \thH(2ch2Hcos 9 - 2sh2H - 1)]2 sin y cos y.
B' = ^2thH(2ch2Hcos9 - 2sh2H- 1) sin j^siny + ZCOSy^).
Finally, substituting the values for A and B' into (5.12) and multiplying the matrix
by ^2thH(2ch2Hcos 9 - 2sh2H- 1) sin y J , we obtain:
n nsin y + i COS y
0
The last matrix represents a bisector of the angle 0, which makes sense since in an
isosceles (hyperbolic or Euclidean) triangle HKO with the base HK, the bisector and the
0A f)siny - ZCOSy
He =
52
median of the angle 0 as well as the perpendicular bisector of HK is the same line, which
is also the Euler line of the triangle.
5.2 When does the Orthocenter Belong to the Euler Line
In Euclidean geometry the orthocenter of a triangle belongs to the Euler line. An
elegant proof of this statement can be found in [2]. Let us just point out that the proof uses
the relationship of similarity of the triangles. Things are different and much more
complicated in non-Euclidean geometry, where there is no distinction between similarity
and d-congruency of the triangles, and so the idea of the Euclidean proof can not be used. (
“In non-Euclidean geometry two d-figures are d-congruent if there is a non-Euclidean
transformation that maps one onto the other”. [1])
Theorem 5.1
If the orthocenter of an isosceles hyperbolic triangle exists then it belongs to the
Euler line.
Proof. Let HKO be an isosceles triangle with HO = KO. Then the perpendicular
bisector of the side HK is also the median and the altitude. Therefore all three points:
circumcenter, centroid and orthocenter belong to the same d -line, and this d -line is the
hyperbolic Euler line of the triangle.
Theorem 5.1 tells us that it is possible for the orthocenter of a hyperbolic triangle
to belong to the Euler line. Although we still do not know, if it is possible for a scalene
hyperbolic triangle to have the named property. To continue our investigation of the
question, let us recall the expression for the orthocenter of a triangle that we found in
Chapter III:
53
[tan# - 7tan20- |y0|2 "1 7 = ------------—U------------- 7o,
l7o|
where yo = th(2K) sin# + i(th(2H) - th(2K) cos#. As in the case of the circumcenter and
the centroid, we want to make same substitutions. We set
yoi = [tan# - ^tan2# - |yo|2 J,
701 =//z(2K)sin# and yo2 = th(2H) - th(2K)cosQ.
Then yo = yoi + zyo2 and
7 =-^u-(7oi+fyo2). l7o|
In order for y to belong to the Euler line, its coordinates have to satisfy the
equation 7/g(y) = 0, or
(513) F *)[z f ](T )=»•
Let us look at the matrix (5.12) that represents the Euler line and then rewrite (5.13) as
follows:
( Z 1 )(5-14)A B' b' A
71
= 0,
where
B' = b\ + ib'2 = (/?02sin# - ao2»z) + sin# - aoi^)
Performing some algebraic transformations of (5.14) we receive:
54
(|712 + 1) A + 2-y-k~(70. b, - 70262) = 0. l7o|2
2 'V jo 9 2Since |7| =—-y and 7io + l7o| = 2710 tan#, from the equation above we obtain IZol
Atan# + 7016', - 70262 = 0.
Substituting A, 701, 702, b\ and 62 for their values, we rewrite the last equation as
follows:
(Poi«o2 - aoi/Wtan# + (/fosin#- ao2»0 th 2/fsin#-
- (/?oi sin# - «oi/u)(^ th 2H- th 2/Tcos#) = 0
The previous equation is equivalent to
(5.15) [(56(2/7) + 56(2/0 cos #)(/6/C- thHcosQ) - thH sh(2K) (1 - cos2#)] +
+ [.y6(2/Q(l - cos2#) - (thK- thHcos#) m]/6(2/Q cos# -
- [(56(2/7) + 56(2/0 cos#) - thH m~\(th(2H) - f6(2/Ocos#)cos# = 0.
Now we need to rewrite the expression on the left hand side of (5.15) as a
polynomial on cos#, and the equation as /’(cos#) = 0, where
(5.16) /’(cos#) = P3 cos3# + P2 cos2# + Pi cos# + Po
We will calculate each coefficient separately:
P3 = -sh(2K)th(2K) sh(2K)th(2K) = 0,
P2 = sh(2H)th(2K) - sh(2K)th(2H)
sh(2H)ch(2H)sh(2K) - sh(2K)sh(2H)ch(2K) ch(2H)ch(2K)
55
P2 = th(2H)th(2K)(ch(2H) - ch(2K)),
Pi = [sh(2K)thK- sh(2H)thH] + [(thHth(2H) - thKth(2K))m\ +
+ [sh(2K)th(2K) - sh(2H)th(2H)]
= -[cA(2flj - ch(2K)] + (ch(2H) + ch(2K) + 1 )(ch(2H) - ch(2K)) ch(2H)ch(2K)
(ch(2H)ch(2K) + V)(ch(2H) - ch(2K)) ch(2H)ch(2K)
(ch(2H) + ch(2K) - 2ch(2H)ch(2K))(ch(2H) - ch(2K)) ch(2H)ch(2K)
Po = sh(2H)thK- sh(2K)thH = thHthK(ch(2H) - ch(2K)).
Now, that all the coefficients are known, we can substitute them into(5.16).
(Before we do that, let us notice that each of the coefficients has a factor
ch(2H) - ch(2K)). We obtain:
P(cos0) = \ch(2H) - ch(2K)][th(2H)th(2K)cos28 +
+ (ch(2H) + ch(2K) - 2ch(2H)ch(2K)) ch(2H)ch(2K) cos 6 + thHthK\,
and therefore P(cos0) = 0 can be written as follows:
[c/z(277) - ch(2K')][sh(2H)sh(2K')cos2d +
+ (ch(2H) + ch(2K) - 2ch(2H)ch(2K))cos3 +
+ thHthKch(2H)ch(2K)\ = 0.
56
The previous equation is equivalent to the union of two equations:
(5.17) [ch(2H) - ch(2K)} = 0
and
(5.18) sh(2Hjsh(2K)cos20 + (ch(2H) + ch(2K) - 2ch(2H)ch(2K))cos6 +
+ thHthKch(2H)ch(2K) = 0
From (5.17) it immediately follows that H = W.This result does not give us anything new,
it was already stated in Theorem 5.1.
Let us now transfer our attention to (5.18), which is the quadratic equation. We
first find the discriminant D :
D = (ch(2H) + ch(2K) - 2ch(2Hjch(2Kf2 - 4sh(2H)sh(2K)thHthKch(2H)ch(2K)
= (ch(2H) - ch(2K))2,
and thus there are two cases to consider.
Case 1:
ch(2H)ch(2K) - ch(2K) sh(2H)sh(2K)
ch(2H) - 1 sh(2H)
ch(2K)sh(2K) = thHcth(2K),
Let us recall the expression (2.5) for cos H from Chapter II. By substituting
cos0 = thHcth(2K) into (2.5), we find that
cosH = thHcth(2K) = cos0,
and thus we are dealing with the isosceles triangle again, only this time the equal sides are
HKandKG.
57
Case 2:
0 ch(2H)ch(2K) - ch(2H) ch(2K) - 1 s6(2H)s6(2£) - sh(2K)
ch(2H)sh(2H) = thKcth(2H).
From (2.6) it follows that when cos0 = thKcth(2H),
cosK = thKcth(2ff) = cos0,
and this time the triangle HKO is isosceles with HK = HO.
Now all the possibilities are exhausted, and we can make a conclusion that will
complete our discussion about the named property of the Euler line.
Theorem 5.2.
Assume that the orthocenter and Euler line of a hyperbolic triangle exist. Then the
hyperbolic Euler line goes through the orthocenter of the triangle if and only if the triangle
is isosceles.
Analytically we can state Theorem 5.2 as follows:
If hyperbolic triangle HKO is such that h = k and 0 is the vertex angle then the
orthocenter belongs to the Euler line if and only if
Qa, if 6 < 45-cos0 > max [0a ,0P] =
Qp, ifh >
58
CHAPTER SIX
SUMMARY
Let us consider Euclidean triangle HOK(e) such that HO = h, KO — k,
/-HOK = 3, and non-Euclidean triangle HOK^ (as in Chapter II), whose vertices
coincide with the vertices of HOK(e). Let also HK(e) = g. Using Cosine Theorem, we can
express g in h, k and 3:
g2 = h2 + k2-2hkcosd.
We can also express g in non-Euclidean terms:
(6.1) g2 = th2H + th2K - 2thHthKcos 3.
Let us now consider Euclidean triangle POQ(e) such that PO = p, QO = q,
/-POQ = 3, and non-Euclidean triangle POQrp) such that its vertices coincide with the
vertices of POQ<e) and PO = 2H, QO = 2K. Let also PQ(e) = s. Using Cosine Theorem,
we can express s in/?, q and 3:
s2 = p2 +q2 - 2pqc.os3
Being expressed in non-Euclidean terms, s will look as follows:
(6.2) s2 = th22H + th22K - 2th2Hth2Kcos 3
The perpendicular bisectors of the sides of a hyperbolic triangle belong to the
same pencil of circles. The circumcenter of the triangle exists if and only if this pencil is
elliptic, which happens if
(6.3) sin2# > |«o|2,
59
where
ao = thH sin 3 + i(thK- thH cos 0).
According to (6.1) we can rewrite the condition (6.3) in Euclidean terms:
sin20 > g2.
When the circumcenter exists it can be found as follows:
(6.4) a = ^sin0- ^sin2# - |«oj2
and the radius R of the circumcircle is
(6.5) R = th~' (£sin0 - 7sin20 - |«o|2 J|«oQ •
The formulas (6.4) and (6.5) above do not work for an isosceles right triangle, in which
case
= thH 2 (1+/),
R = th~' thH
a
The three medians of a hyperbolic triangle are always concurrent. We calculated
the point of their concurrence /J :
P=[m- Jm2-\Po\2 ~^P7
where
Po = (sh(2H) + sh(2K) cos0) + ish(2K) sin0,
and called P the centroid of a hyperbolic triangle by analogy with Euclidean geometry,
although we never checked whether the point of concurrence of the medians is really the
60
centroid of the hyperbolic triangle. If in A HOK we connect point P with its vertices, we
obtain three hyperbolic triangles. We will say that P is the centroid of the hyperbolic
triangle if the areas of the triangles HOfi, KO ft, and HKP are equal.
The three altitudes of a hyperbolic triangle belong to the same pencil of circles,
which is elliptic if
2pq cos3# - (p2 + q2 + 1) cos2# + 1 > 0.
This condition is equivalent to the following :
(6.6) tan2# > |yo |2,
where
yo = th(2K) sin# + i(th(2H) - th(2K) cos#).
It follows from (6.2) that we can right (6.6) as:
tan2# > s2.
In the case when the pencil is elliptic, the altitudes are concurrent at the
orthocenter y of the triangle, which we can find as follows:
y = pan# ± .Jtan2# - |y0|2 ]/o>
choosing the negative square root for the acute # and positive for the obtuse(in the right
triangle y = 0).
The hyperbolic Euler line exists if and only if the condition (6.3) holds. If the
Euler line exists, it can be represended by the Hermitian matrix He :
61
he = Im(aoPo) i{ma<j - /?0 sin#)-i(mao - j#osin#) Im(a0/?0)
If 0 is such that
1 - cos4# > |<x0|2 + |7o|2,
then both the circumcenter and the orthocenter exist.
As a result of this study we conclude the following theorem about the property of
the Euler line:
Assume that the orthocenter and Euler line of a hyperbolic triangle exist. Then the
hyperbolic Euler line goes through the orthocenter of the triangle if and only if the triangle
is isosceles.
Analytically we can state this theorem as follows:
If hyperbolic triangle HKO is such that h = k and 0 is the vertex angle then the
orthocenter belongs to the Euler line if and only if
9a, if6 < ^-cos # > max [#a , 0p] = .
0p, iih>
62
REFERENCES
[1] Brannan, David A.; Esplen, Matthew F.; and Gray, Jeremy J. Geometry, Cambridge Press, University Cambridge, 1999.
[2] Coxeter, H. S. M. Introduction to Geometry, 2nd Edition, John Wiley & Sons, New York, 1969.
[3] Coxeter, H. S. M. Non-Euclidean Geometry, Mathematical Expositions, No. 2, 4th Edition, University of Toronto Press, Toronto, 1961.
[4] Greenberg, Marvin Jay. Euclidean and Non-Euclidean Geometries: Development and History, 3rd Edition, W.H. Freeman & Company, New York, 1993.
[5] Isaacs, I. Martin. Geometry for College Students, Brooks/Cole Thomson Learning, Pacific Grove, CA, 2001.
[6] Martin, George E. The Foundations of Geometry and the Non-Euclidean Plane, Springer-Verlag, New York, 1975.
[7] Rosenfeld, B. A. A History of Non-Euclidean Geometry: Evolution of the Concept of a Geometric Space, Springer-Verlag, New York, 1988.
[8] Schwerdtfeger, Hans. Geometry of Complex Numbers: Circle Geometry, Moebius Transformation, Non-Euclidean Geometry, Dover Publications, New York, 1979.
[9] Smogorzhevsky, A. S. Lobachevskian Geometry, translated from the Russian by V. Kisin, Mir Publishers, Moscow, 1976.
[10] Trudeau, Richard J. The Non-Euclidean Revolution, Birkhauser, Boston, 1987.
[11] Wolfe, Harold E. Introduction to Non-Euclidean Geometry, Holt, Rinehart & Winston, New York, 1945.
[12] Bronstein, I. N. and Semendyayev, K. A Handbook of Mathematics, (in Russian), Nauka Press, Moscow, 1980.
[13] http://www.maths.gla.ac.uk/~wws/cabripages/hyperbolic/ceva.html .
63
Recommended